Public comments help officials hear from people affected by proposed policies. But when AI can produce fluent, varied submissions without making their origin obvious, reviewers may struggle to tell whether a comment reflects a person’s view. Deb Roy, a professor of media arts and sciences at MIT and director of the MIT Center for Constructive Communication (CCC), says the issue reaches beyond authorship: someone must be accountable for the words used in civic debate.
When polished text obscures who is speaking
Earlier efforts to flood public feedback channels with fake submissions often left clues, such as repeated wording or templates. Roy points to the debate over repealing net neutrality rules: in 2017, the Federal Communications Commission received more than 22 million public comments, and the New York attorney general later found that nearly 18 million were fake. As Roy describes it, duplicated phrasing made the fraud detectable.
AI-generated comments can be harder to identify because they may read fluently and differ from one another. That leaves officials with a difficult question: does a submission represent a person who formed and supports the view it expresses, or is it text without an identifiable speaker behind it?
Roy says a large volume of unaccountable comments can weaken the value of the whole channel. If reviewers become less confident that submissions are genuine, they may give public feedback less weight. People who take time to participate may then feel that officials are not listening, further damaging trust in the process.
Why public input needs a responsible speaker
Roy describes public comment as a pillar of democracy. Before a rule is imposed, people need a chance to express their views and to feel heard. When that exchange loses credibility, the process risks losing some of its legitimacy in the eyes of the public.
His concern is not limited to whether a language model understands what it writes. That question remains debated, Roy says. Even if one assumes a system can understand language, he argues, the system itself cannot be answerable for its words: it has no personal stake in what follows from them.
For Roy, “standing” offers a useful way to think about the human role. He is borrowing the idea informally, rather than proposing a new formal rule. In an open town meeting, for example, a speaker is generally expected to have a connection to the town and a stake in what is decided. That connection helps others understand why the person is speaking and who may be held responsible for the contribution.
Using AI while keeping people in the loop
The MIT Center for Constructive Communication and Cortico, a closely affiliated nonprofit Roy co-founded, have helped organize group discussions in civic meetings, schools, deliberative assemblies, and private organizations. Their work uses technology, including AI, to identify themes across comments and support the discussion process.
In that work, participants can review and override technology-based outputs at key steps. The design keeps people involved in assessing how the group’s comments are summarized. Roy’s stated principle is that a person should ultimately be responsible for each public comment.
That approach points to a practical distinction: AI may assist with organizing or expressing information, while a person remains answerable for submitting a view in a civic setting. The tool can help process words, but responsibility still needs to connect those words to someone with a stake in the conversation.
Keeping civic feedback credible
Public consultation depends on more than collecting a large number of submissions. Officials and participants also need confidence that the input reflects people who are willing to stand behind what they say. If realistic AI text makes that harder to judge, simply counting comments may offer less insight into public views.
Roy’s argument is that accountability should remain part of civic communication even as the tools for producing text change. Giving participants a chance to check technology-assisted summaries, and making clear that a person is responsible for a submission, can help preserve the link between public words and public voices.